MolProbity

The clashscore quantifies the number of unfavorable steric overlaps per 1,000 atoms in a protein model. It is calculated by counting pairs of non-bonded atoms that are separated by less than 0.4 Å of the sum of their van der Waals radii.

• A clashscore of 0 indicates a physically perfect model with no steric overlaps
• Scores below 5 are typical for high-resolution crystal structures
• Scores above 40 indicate severe modeling errors requiring refinement
• The metric is normalized by model size, enabling direct comparison between structures of different sizes

The Ramachandran plot maps the backbone dihedral angles phi (φ) and psi (ψ) for each residue, identifying conformations that fall into energetically disallowed regions. MolProbity classifies residues into three categories:

• Favored: Residues occupying the most energetically favorable regions (target: > 98%)
• Allowed: Residues in less favorable but still permissible regions
• Outliers: Residues in sterically impossible conformations, indicating local backbone errors

Glycine and proline are evaluated against residue-specific distributions due to their unique conformational properties.

Rotamer analysis evaluates the discrete conformational states of amino acid side chains against a curated library of statistically preferred rotamers derived from high-resolution crystal structures.

• Each side chain is assigned a rotamericity score based on how closely it matches the nearest ideal rotamer
• Outlier rotamers indicate strained or physically improbable side-chain packing
• The analysis accounts for backbone-dependent preferences, where the local backbone conformation influences side-chain positioning
• Systematic rotamer outliers often reveal errors in sequence registration or local backbone tracing

Cβ deviation measures the displacement of the beta-carbon atom from its ideal geometric position relative to the backbone atoms. This metric is particularly sensitive to errors in backbone tracing and sequence registration.

• Large Cβ deviations (> 0.25 Å) often indicate an incorrect residue identity at that position
• The metric is especially useful for validating mutated or designed proteins where side-chain geometry may be strained
• Combined with Ramachandran outliers, Cβ deviations provide strong evidence for local model rebuilding requirements

The MolProbity score is a composite quality metric that combines clashscore, Ramachandran outliers, and rotamer outliers into a single normalized value. It is calibrated to approximate the resolution-dependent quality expected from X-ray crystallography.

• A score of 1.0 represents the expected quality of a well-refined structure at its reported resolution
• Scores significantly above 2.0 indicate systematic problems requiring attention
• The score enables cross-model ranking in CASP competitions and structure prediction benchmarks
• It provides a unified, intuitive measure for non-specialist assessment of model quality

MolProbity performs all-atom hydrogen bond analysis by adding explicit hydrogen atoms and evaluating the geometry of potential hydrogen bonds. This analysis identifies:

• Unsatisfied donors and acceptors buried in the protein core that lack appropriate hydrogen bonding partners
• Unfavorable electrostatic interactions where like charges are juxtaposed without compensating interactions
• The analysis uses quantum-mechanically optimized hydrogen positions rather than idealized geometry
• Unsatisfied buried polar atoms are strong indicators of local folding errors or missing water molecules